Our analysis suggests several ways in which markets may fail to maximize social welfare, leading to potential improvements from public policy. Below, we describe fi ve types of market failures: wage externalities, information externalities, training externalities, coordination problems and liquidity constraints. These failures lead to the possibility that government inter- vention could increase social welfare. That is, a dollar of public spending might lead to more than a dollar’s worth of benefi ts. Our research suggests that at least the fi rst two forms of market failure may be operating, and that coordination problems may be present as well. It is also possible that government intervention could reduce social welfare. In this section, we examine these potential effects of government policy.
We will also examine other research on how the MEP program has performed in these areas, and present some new information from our survey on the extent of MEP use among these smaller fi rms. First, our fi ndings on urban wages suggest a potential ‘wage externality’ for highly productive urban fi rms. Firms that pay a higher wage are advantageous to workers. However, profi t-maximizing owners will not take into account the benefi ts of higher wages that accrue solely to workers. Luria (1996b) has found that certain production practices, such as capital intensity and distinctive products, are associated with higher wages.
Shane 01 intro 55
Shane 01 intro 55 19/9/05 09:28:3619/9/05 09:28:36
We also found that fi rms in urban areas are more productive than are other fi rms, and that most of these productivity benefi ts are captured by wage earners. Assuming the correlation between urban location and productivity can be interpreted causally, since fi rm owners do not benefi t much from the increased productivity of urban locations, they are likely to undervalue the urban productivity advantage, leading to ineffi ciently low urban employment. In economics language, the urban productivity advantage is largely an ‘externality’, a benefi t not taken into account by those who make fi rm location decisions. Policies that benefi t urban fi rms can remedy some of this ineffi ciency. That is, a dollar of tax money spent in some way on an urban fi rm has the potential to return more than a dollar of benefi ts to society, in the form of a rise in productivity that is shared among fi rm owners, workers and consumers.
Second, our fi nding that single-plant fi rms benefi t from networking with other fi rms implies potential market failure. Information exchange is subject to many market failures. A key issue is that knowledge is ‘expensive to produce, but cheap to reproduce’ (Varian and Shapiro, 1998). That is, if one fi rm knows something, it is ineffi cient for another fi rm to discover that same thing for itself. Yet it is usually not profi table for a fi rm to give away its knowledge for nothing.4 Therefore spending a dollar of tax money on knowledge diffusion may yield more than a dollar of benefi ts by avoiding duplication of discovery.
The discovery process is particularly expensive and diffi cult if changes are complementary (for example, if two modifications made together yield greater performance gains than the sum of the two modifi cations made separately). For example, adopting Toyota-inspired ‘lean production techniques’ leads to higher quality and lower inventory, but only if inventory reduction and quality control are coupled (MacDuffi e, 1995). Each of these initiatives is complex, but fi rms that do inventory reduction without quality control are likely to be plagued by supply shortages.
A third problem is training externalities. In our data, we find that employees work for several fi rms during their careers. As Becker (1975) has pointed out, if workers are mobile, profi t-maximizing fi rms will provide less than the socially optimal amount of general training, because they fear that they will not get the full benefi t of their training expenditure because the trained employees will be hired away by other fi rms.
A fourth problem is liquidity constraints. Adopting the production processes that lead to high wages and high value-added requires capital and product development capability. These upgrading activities require fairly large upfront expenditures. Since many of these expenditures do not result in a tangible asset, banks are usually not willing to lend money to help fi nance them.
A fi nal problem is coordination. Most component manufacturers serve a number of customers. We found that the typical fi rm gets only 30 per cent of its sales from its largest customer. If customers can rely on suppliers to provide timely delivery and high-quality products, they can adopt more effi cient production processes. For example, they can eliminate receiving inspection and expediters. But if suppliers do not all invest in these activities, customers cannot risk running low-inventory production processes.
Our fi ndings on urban agglomeration economies may also imply the potential for coordination failure, although this is unclear without further investigation of the causes of the agglomeration economies. If these economies depend on the simultaneous presence of many different types of fi rms and institutions, fi rms may create signifi cant externalities by locating in urban areas. However our fi nding that cluster economies do not appear to be important for this manufacturing sector does lower the chance that these externalities are taking place within manufacturing; they are likely present in other supporting institutions or in urban infrastructure. Further research is necessary here.
The above processes suggest ways that government intervention could improve welfare, but there also are a variety of ways in which it could reduce welfare. It is possible that programs such as MEP might be welfare-reducing:
by promoting capabilities that the market does not want, by subsidizing fi rms to do things they would otherwise pay for themselves, and by allowing low-wage fi rms to obtain skills they would otherwise have to pay higher wages to get.
As we have seen, the trend in component manufacturing appears to go against the types of fi rms that do relatively better in urban areas, and those that pay high wages. Above we considered the possibility that market failures are leading fi rms to underinvest (from a social point of view) in training, wages and capital. But it is also possible that public money spent on capability improvement does not have an acceptable rate of return even when these externalities are considered.
A second possibility is that subsidized assistance merely substitutes for expenditures on training and consulting services that fi rms (rather than taxpayers) would otherwise make themselves. Even worse is the possibility that subsidized assistance helps drive out more responsible competitors who develop capabilities on their own. In this scenario, the subsidy would be a negative externality to ‘good’ fi rms (Luria, 1996b).
To summarize briefly, we suggest that policy would be likely to be welfare-improving if it (a) promotes the growth of fi rms that are urban and high-wage, (b) provides fi rms with information about techniques that may be useful to them, and (c) helps suppliers and customers coordinate on adopting complementary modern manufacturing methods. It would be
Shane 01 intro 57
Shane 01 intro 57 19/9/05 09:28:3619/9/05 09:28:36
welfare-reducing if (a) fi rms were not able fi nd a use for capabilities gained through MEP training, (b) it duplicated services already available on the private market, or (c) it primarily benefi ted low-wage fi rms (and did not lead to higher wages).
To move from theory to more specifi c policy options, there are three kinds of policies typically recommended for improving manufacturing.
The fi rst set is essentially transfers, such as tax reduction, from some other group toward manufacturers. Despite their strong backing by groups such as the National Association of Manufacturers, these policies typically do not infl uence plant location, let alone increase national welfare. The reason is that (a) taxes are a small part of manufacturers’ costs and (b) when taxes fall, so do public services that manufacturers depend on, such as roads, police protection, education and so on (see the review by Lynch, 2004). The second set of policies tries to improve the supply of high-quality inputs, by subsidizing such activities as training, R&D and capital. Many of these policies have positive effects. The third set attempts to improve the way that the inputs are mixed together. That is, these policies attempt to change fi rms’ production functions.
The Manufacturing Extension Partnership has tried to implement the second and third types of policies. The MEP program was loosely modeled on the agricultural extension program, although the rate of subsidy was much lower (Shapira, 1995). The MEP was set up in 1989 and is administered by the National Institute of Standards and Technology (NIST). Federal support for manufacturing extension activities grew from $6.1 million in 1988 to $138.4 million in 1995, before dropping to $106.6 million in recent years. Federal support to individual centers must be at least matched by state and local sources. Jarmin (1999) describes the activities of the centers:
Manufacturing extension centers provide technical and business assistance to small and medium-sized manufacturers, much as agricultural extension agents do for farmers. This assistance often consists of providing ‘off-the-shelf’ solutions to technical problems. Examples might include helping a plant install a CAD/CAM system or switching to newer, lower cost, higher performance materials. Manufac- turing extension centers can also channel more recent innovations generated in government and university laboratories to SMEs that lack access to such informa- tion. Besides helping plants adopt modern manufacturing technologies, most centers also offer business, marketing, and other ‘softer’ types of assistance.
How well have MEPs done in improving fi rm productivity? Jarmin (1999) conducted a careful study of the early years of the MEP program that is superior to what is possible with our data. Using the Census Bureau’s Lon- gitudinal Research Database, he estimated that productivity at MEP client
fi rms rose 3.4–16 per cent more between 1987 and 1992, compared to prod- uctivity at non-client fi rms (depending on the method of estimation).
Jarmin’s study takes a novel approach to the problem that participation in the program is not random. Firms who are either more productive than average (and therefore more aggressive) may be more likely to seek out the program, or fi rms who are less productive than average (and therefore more desperate for help) may be more likely to use the program. In either case, the estimates of the effect of the MEP ‘treatment’ will be biased. Jarmin corrected for this bias by observing that fi rms that are closer to an MEP are more likely to use it. His statistical method thus implicitly compares the productivity of two fi rms that are identical except that one is close to an MEP center and one is not.
Jarmin does not attempt to compare these benefi ts to the costs of the program. However a rough estimate is possible using data contained in Jarmin (1999) and in Shapira (2003). Project costs charged to the client average $67 787; Shapira says that these are typically one-third of total costs (one-third of the total comes from the federal government and one third from the state match), so total costs would be $191 361. If the increase in value-added is conservatively estimated at 3.4 per cent (Jarmin’s lowest estimate), the average fi rm had $306 340 more value-added as a result of the program than it would have had otherwise. If we assume that the gain compared to non-clients dissipates over time, so that after fi ve years value-added is the same as at non-clients, the payback period is 1.6 years – not a bad investment. If the productivity advantage continues, the investment is even more productive.
This result suggests that total benefi ts to society outweigh the costs. This fi nding, plus overwhelming reports by participants that the services provided were useful (Shapira, 2003), suggests that MEP is developing capabilities that have market applicability. However, the case for MEP intervention in the previous section relied heavily on the existence of externalities – benefi ts that fl ow to people other than those who make decisions for the fi rm. Benefi ts that fl ow to workers could relatively easily be measured by comparing wages in treatment and control groups. Using a different methodology, Luria (1997) did this comparison, and found no difference.
The benefi ts to customers would be hard to measure. To the extent that the MEP program increases the supply of qualifi ed suppliers, component prices will fall. This effect would cause measured productivity (dollar value of output/ labor hour) to fall, suggesting that Jarmin’s estimate of total productivity increase is conservative.
Jarmin also provides data on who participates in MEP programs. Firms are much more likely to participate if an MEP center is geographically close to them. Since centers are more likely to be in urban areas, this benefi ts urban fi rms.
Shane 01 intro 59
Shane 01 intro 59 19/9/05 09:28:3619/9/05 09:28:36
Small fi rms benefi t more from MEP programs, but participate less. We found that most of the small fi rms we surveyed did not appear to take advantage of MEP assistance. Only 6 per cent of these small manufactur- ing fi rms reported receiving external assistance from a publicly supported manufacturing extension center at any time in the previous three years.
Since these centers are especially aimed at small manufacturing fi rms, this is somewhat surprising (Shapira, 2003). There may be some recall error here, but use of the centers does not appear to be widespread in our sample.
Why do small fi rms make so little use of this resource? MEPs often teach courses piecemeal, without offering an overall improvement plan to the fi rm. Even if such a plan is offered, liquidity constraints and lack of organizational slack make it diffi cult for small fi rms to undertake a sustained program of improvement (Helper and Kiehl, 2004). Cutbacks in federal funding since the time of Jarmin’s study have caused several MEPs, such as CAMP, the MEP in Northeast Ohio, to focus efforts even more on large fi rms which often require less subsidy (interview with CAMP president Stephen J. Gage, January 2004). On the other hand, the Pennsylvania MEPs are serving disproportionately small fi rms (Deloitte and Touche, 2004).
Most MEPs focus their work on either remedying information problems or coordination failures. They offer a wide variety of activities, and the programs emphasized by centers vary even within states. (For example, in Pennsylvania, some centers focus almost exclusively on teaching lean production, while others do very little on lean production and much more on introducing new technology.) However MEPs could do much more to remedy coordination failures by organizing their work around value chains rather than focusing on individual fi rms.
An exception is the consortial model of supply chain modernization used by the Wisconsin MEP. It set up the Wisconsin Manufacturers’ Development Consortium (WMDC), which provides a single venue for training providers and trains suppliers in general (rather than OEM-specifi c) competencies, and promotes mutual learning by harmonizing supplier certifi cation and encouraging cross-supplier communication. This framework meets diverse supplier needs through multiple institutional supports. For example, major improvements at formerly struggling suppliers resulted from a mix of WMDC supplier training, OEM-led (project-based) development and internal initiatives at suppliers (Whitford and Zeitlin, 2004).
There is limited evidence that MEPs are keeping alive ‘bad’ competitors.
Deloitte and Touche (2004) found that the credit scores of Pennsylvania MEP clients are worse than those of non-clients. Deloitte and Touche argue that this is a positive fi nding, since it means that the MEPs are not cream skimming. On the other hand, low credit scores in the absence of some market failure may be an indication that MEPs are aiding ineffi cient fi rms. Jarmin
(1999) fi nds that the typical MEP user is a fast-growing, low-productivity fi rm. These fi rms could either be fi rms that have a distinctive product but are ineffi cient producers, or are low-cost, ‘commodity’ fi rms (Luria and Wiarda, 1996). If they are ineffi cient producers, Luria and Wiarda found that MEP customers improve faster than non-MEP customers in adopting most technologies, except information technologies. In his review of this literature, Shapira (2003) concludes that the studies ‘suggest that not all desired policy outcomes can be achieved simultaneously’.